Bézier Curves and Splines

Transcription

1 CS-C3100 Computer Graphics Bézier Curves and Splines Majority of slides from Frédo Durand vectorportal.com CS-C3100 Fall 2017 Lehtinen

2 Before We Begin Anything on your mind concerning Assignment 1? CS-C3100 Fall 2017 Lehtinen 2

3 Today Curves in 2D Useful in their own right Provides basis for surface modeling (this Thursday) CS-C3100 Fall 2017 Lehtinen Guilherme Marconi

4 Modeling Curves in 2D How? CS-C3100 Fall 2017 Lehtinen 4

5 Modeling Curves in 2D Polylines Sequence of vertices connected by straight line segments Useful, but not for smooth curves Very easy! This is the representation that usually gets drawn in the end (smooth curves converted into these) Smooth curves How do we specify them? A little harder (but not too much) CS-C3100 Fall 2017 Lehtinen 5

6 Splines A type of smooth curve in the plane or in 3D Many uses 2D Illustration (e.g. Adobe Illustrator) Fonts 3D Modeling Animation: Trajectories In general: Interpolation and approximation ACM 1987 Principles of traditional animation applied CS-C3100 Fall 2017 Lehtinen to 3D computer animation 6

7 How Many Dimensions? CS-C3100 Fall 2017 Lehtinen 7

8 How Many Dimensions? This curve lies on the 2D plane, but is itself 1D. CS-C3100 Fall 2017 Lehtinen 8

9 How Many Dimensions? This curve lies on the 2D plane, but is itself 1D. CS-C3100 Fall 2017 Lehtinen You can just as well define 1D curves in 3D space. 9

10 Two Definitions of a Curve 1.A continuous 1D point set on the plane or space 2.A mapping from an interval S onto the plane That is, P(t) is the point of the curve at parameter t P : R S R 2, P(t) = x(t) y(t) Big differences It s easy to generate points on the curve from the 2nd The second definition can describe trajectories, the speed at which we move on the curve CS-C3100 Fall 2017 Lehtinen 10

11 Example of the First Definition? A continuous 1D point set on the plane or space A mapping from an interval S onto the plane That is, P(t) is the point of the curve at parameter t P : R S R 2, P(t) = x(t) y(t) CS-C3100 Fall 2017 Lehtinen 11

12 Example of the First Definition? A continuous 1D point set on the plane or space A mapping from an interval S onto the plane That is, P(t) is the point of the curve at parameter t P : R S R 2, P(t) = x(t) y(t) An algebraic curve, e.g. intersection of x 2 + y 2 + z 2 =1 and z =0 It s a circle on the xy plane :) CS-C3100 Fall 2017 Lehtinen 12

13 General Principle for Modeling User specifies control points We ll interpolate the control points by a smooth curve The curve is completely determined by the control points. I.e., we need an unambiguous rule that gives us the curve points given the control points. CS-C3100 Fall 2017 Lehtinen 13

14 Physical Splines See CS-C3100 Fall 2017 Lehtinen 14

15 Two Points of View Approximation/interpolation We have data points, how can we interpolate? Important in lots of applications, both graphics and non-graphics User interface/modeling What is an easy way to specify a smooth curve? Our main perspective today. celshader.com vectorportal.com CS-C3100 Fall 2017 Lehtinen 15

16 Splines Specified by a few control points Good for UI Good for storage Results in a smooth parametric curve P(t) Just means that we specify x(t) and y(t) In practice low-order polynomials chained together Convenient for animation, where t is time Convenient for tessellation because we can discretize t and approximate the curve with a polyline CS-C3100 Fall 2017 Lehtinen 16

17 Tessellation tn t1 t2 t0 CS-C3100 Fall 2017 Lehtinen 17

18 Tessellation tn t1 t2 t0 It s clear that adding more points will get us closer to the curve. CS-C3100 Fall 2017 Lehtinen 18

19 Interpolation vs. Approximation Interpolation Goes through all specified points Sounds more logical Approximation Does not go through all points Interpolation Approximation CS-C3100 Fall 2017 Lehtinen 19

20 Interpolation vs. Approximation Interpolation Goes through all specified points Sounds more logical But can be more unstable, ringing Approximation Does not go through all points Turns out to be convenient Interpolation In practice, we ll do something in between. Approximation CS-C3100 Fall 2017 Lehtinen 20

21 Questions? CS-C3100 Fall 2017 Lehtinen 21

22 Cubic Bézier Curve User specifies 4 control points P1... P4 Curve goes through (interpolates) the ends P1, P4 Approximates the two other ones Cubic polynomial CS-C3100 Fall 2017 Lehtinen 22

23 Cubic Bézier Curve P(t) = (1-t) 3 P 1 + 3t(1-t) 2 P 2 + 3t 2 (1-t) P 3 + t 3 P 4 That is, x(t) = (1 t) 3 x 1 + 3t(1 t) 2 x 2 + 3t 2 (1 t) x 3 + t 3 x 4 y(t) = (1 t) 3 y 1 + 3t(1 t) 2 y 2 + 3t 2 (1 t) y 3 + t 3 y 4 CS-C3100 Fall 2017 Lehtinen 23

24 Cubic Bézier Curve P(t) = (1-t) 3 P 1 + 3t(1-t) 2 P 2 + 3t 2 (1-t) P 3 + t 3 P 4 Scalar polynomials That is, Vectors with coordinates of control points x(t) = (1 t) 3 x 1 + 3t(1 t) 2 x 2 + 3t 2 (1 t) x 3 + t 3 x 4 y(t) = (1 t) 3 y 1 + 3t(1 t) 2 y 2 + 3t 2 (1 t) y 3 + t 3 y 4 CS-C3100 Fall 2017 Lehtinen 24

25 Cubic Bézier Curve P(t) = (1-t) 3 P 1 + 3t(1-t) 2 P 2 + 3t 2 (1-t) P 3 + t 3 P 4 Verify what happens for t=0 and t=1 CS-C3100 Fall 2017 Lehtinen 25

26 Cubic Bézier Curve 4 control points Curve passes through first & last control point Curve is tangent at P 1 to (P 1 -P 2 ) and at P 4 to (P 4 -P 3 ) A Bézier curve is bounded by the convex hull of its control points. CS-C3100 Fall 2017 Lehtinen 26

27 Cubic Bézier Curve 4 control points Curve passes through first & last control point Curve is tangent at P 1 to (P 1 -P 2 ) and at P 4 to (P 4 -P 3 ) A Bézier curve is bounded by the convex hull of its control points. CS-C3100 Fall 2017 Lehtinen 27

28 Bernstein Polynomials For cubic: B 1 (t)=(1-t) 3 B 2 (t)=3t(1-t) 2 B 3 (t)=3t 2 (1-t) B 4 (t)=t 3 (careful with indices, many authors start at 0) But defined for any degree CS-C3100 Fall 2017 Lehtinen 28

29 Properties of Bernstein polynomials 0 for all 0 t 1 Sum to 1 for every t called partition of unity (These two together are the reason why Bézier curves lie within convex hull) Only B 1 is non-zero at 0 B 1 + B 2 + B 3 + B 4 Bezier interpolates P 1 Same for B 4 and P 4 for t=1 CS-C3100 Fall 2017 Lehtinen 29

30 Bézier Curves P(t) = P 1 B 1 (t) + P 2 B 2 (t) + P 3 B 3 (t) + P 4 B 4 (t) P i are 2D control points (xi, yi) For each t, the point P(t) on a Bézier curve is a linear combination of the control points with weights given by the Bernstein polynomials at t CS-C3100 Fall 2017 Lehtinen 30

31 Interpretation as Influence Each B i specifies the influence of P i First, P 1 is the most influential point, then P 2, P 3, and P 4 P 2 and P 3 never have full influence Not interpolated! CS-C3100 Fall 2017 Lehtinen 31

32 Questions? CS-C3100 Fall 2017 Lehtinen 32

33 What s with the Formula? Explanation 1: Magic! Explanation 2: It s a linear combination of basis polynomials Let s study this in 1D using curves y=f(t) CS-C3100 Fall 2017 Lehtinen 33

34 Usual Vector Spaces In 3D, each vector has three components x, y, z But geometrically, each vector is actually the sum v = x i + y j + z k i, j, k are basis vectors j k i Vector addition: just add components Scalar multiplication: just multiply components CS-C3100 Fall 2017 Lehtinen 34

35 Polynomials as a Vector Space Polynomials Can be added: just add the coefficients y(t) =a 0 + a 1 t + a 2 t a n t n (y + z)(t) =(a 0 + b 0 )+(a 1 + b 1 )t + (a 2 + b 2 )t (a n + b n )t n Can be multiplied by a scalar: multiply the coefficients s y(t) = (s a 0 )+(s a 1 )t +(s a 2 )t (s a n )t n CS-C3100 Fall 2017 Lehtinen 35

36 Polynomials as a Vector Space In 3D, each vector has three components x, y, z But geometrically, each vector is actually the sum v = x i + y j + z k i, j, k are basis vectors j k i In the polynomial vector space, {1, t,..., t n } are the basis vectors, a0, a1,..., an are the components y(t) =a 0 + a 1 t + a 2 t a n t n CS-C3100 Fall 2017 Lehtinen 36

37 Polynomials as a Vector Space Polynomials are like vectors; you can add them and scale them, and the result is still a polynomial. Connections The flavor of math that is concerned with such things is called functional analysis. Many familiar concepts, such as orthogonality, carry over from ordinary n-d vectors to functions! Trivializes many difficult things like Fourier series by giving geometric analogue. Many other uses in graphics, e.g. finite element global illumination algorithms such as radiosity. THEVERYMANY CS-C3100 Fall 2017 Lehtinen 37

38 Questions? CS-C3100 Fall 2017 Lehtinen 38

39 Subset of Polynomials: Cubic y(t) =a 0 + a 1 t + a 2 t 2 + a 3 t 3 Closed under addition & scalar multiplication Means the result is still a cubic polynomial (verify!) Not under general multiplication: t 3 * t 3 = t 6! This means it is also a vector space, a 4D subspace of the full space of polynomials How many dimensions does the full space have? The x and y coordinates of cubic Bézier curves belong to this subspace as functions of t. CS-C3100 Fall 2017 Lehtinen 39

40 Basis for Cubic Polynomials More precisely: What s a basis? v = x i + y j + z k In 3D A set of atomic vectors Called basis vectors Linear combinations of basis vectors span the space Linearly independent Means that no basis vector can be obtained from the others by linear combination Example: i, j, i+j is not a basis (missing k direction!) j k i CS-C3100 Fall 2017 Lehtinen 40

41 Canonical Basis for Cubics 1 {1, t, t 2,t 3 } Any cubic polynomial is a linear combination of these: 1 t t 2 t a 0 + a 1 t + a 2 t 2 + a 3 t 3 = a 0 * 1+ a 1 * t + a 2 * t 2 + a 3 * t 3 They are linearly independent Means you can t write any of the four monomials as a a linear combination of the others. (You can try.) CS-C3100 Fall 2017 Lehtinen 41

42 Different basis 2D examples For example: {1, 1+t, 1+t+t 2, 1+t-t 2 +t 3 } {t 3, t 3 +t 2, t 3 +t, t 3 +1} These can all be obtained from 1, t, t 2,t 3 by linear combination Just like all bases for Euclidean space can be obtained by linear combinations of the canonical i, j, Infinite number of possibilities, just like you have an infinite number of bases to span R 2 CS-C3100 Fall 2017 Lehtinen 42

43 Why we bother: CS-C3100 Fall 2017 Lehtinen 43

44 Matrix-Vector Notation For Polynomials For example: 1, 1+t, 1+t+t 2, 1+t-t 2 +t 3 t 3, t 3 +t 2, t 3 +t, t 3 +1 These relationships hold for each value of t = t t t 3 t 3 t 3 + t 2 t 3 + t = t t 2 t t t 1+t + t 2 1+t t 2 + t 3 Change-of-basis matrix Canonical monomial basis CS-C3100 Fall 2017 Lehtinen 44

45 Matrix-Vector Notation For Polynomials For example: 1, 1+t, 1+t+t 2, 1+t-t 2 +t 3 t 3, t 3 +t 2, t 3 +t, t 3 +1 Not any matrix will do! If it s singular, the basis set will be linearly dependent, i.e., redundant. = t t t 3 t 3 t 3 + t 2 t 3 + t = t t 2 t t t 1+t + t 2 1+t t 2 + t 3 Change-of-basis matrix Canonical monomial basis CS-C3100 Fall 2017 Lehtinen 45

46 Matrix Form of Bernstein Cubic Bernstein: B 1 (t)=(1-t) 3 B 2 (t)=3t(1-t) 2 B 3 (t)=3t 2 (1-t) B 4 (t)=t 3 B 1 (t) B 2 (t) B 3 (t) B 4 (t) = Expand these out and collect powers of t. The coefficients are the entries in the matrix B! B CS-C3100 Fall 2017 Lehtinen 46 1 t t 2 t 3

47 More Matrix-Vector Notation Remember: P(t) = P 1 B 1 (t) + P 2 B 2 (t) + P 3 B 3 (t) + P 4 B 4 (t) is a linear combination of control points or, in matrix-vector notation P (t) = x(t) y(t) point on curve (2x1 vector) = x 1 x 2 x 3 x 4 y 1 y 2 y 3 y 4 matrix of control points (2 x 4) CS-C3100 Fall 2017 Lehtinen Bernstein polynomials (4x1 vector) B 1 (t) B 2 (t) B 3 (t) B 4 (t) 47

48 More Matrix-Vector Notation Remember: B 1 (t) B 2 (t) B 3 (t) B 4 (t) P (t) = P (t) = = x(t) y(t) point on curve (2x1 vector) B t t i=1 3 P i B i (t) t 2 = matrix of control points (2 x 4) CS-C3100 Fall 2017 Lehtinen 4 <= Flasback from two xi slides ago, B y i (t) let s combine i with below: i=1 = x 1 x 2 x 3 x 4 y 1 y 2 y 3 y 4 Bernstein polynomials (4x1 vector) B 1 (t) B 2 (t) B 3 (t) B 4 (t) 48

49 Phase 3: Profit Combined, we get cubic Bézier in matrix notation point on curve (2x1 vector) P(t) = x(t) y(t) x 1 x 2 x 3 x 4 y 1 y 2 y 3 y 4 Geometry matrix of control points P1..P4 (2 x 4) = Spline matrix (Bernstein) CS-C3100 Fall 2017 Lehtinen Canonical monomial basis 1 t t 2 t 3 49

50 General Spline Formulation Geometry: control points coordinates assembled into a matrix (P 1, P 2,, P n+1 ) Spline matrix: defines the type of spline Bernstein for Bézier Power basis: the monomials (1, t,..., t n ) Advantage of general formulation Compact expression Easy to convert between types of splines Dimensionality (plane or space) doesn t really matter CS-C3100 Fall 2017 Lehtinen 50

51 What can we do with this? Cubic polynomials form a vector space. Bernstein basis defines Bézier curves. Can do other bases as well: Catmull-Rom splines interpolate all the control points, for instance B CR = B A CS-C3100 Fall 2017 Lehtinen 51

52 3D Spline has 3D control points The 2D control points can be replaced by 3D points this yields space curves. All the above math stays the same except with the addition of 3rd row to control point matrix (z coords) In fact, can do homogeneous coordinates as well! CS-C3100 Fall 2017 Lehtinen 52

53 Easy Brain Teaser Can you derive formulas for conversion between two different cubic spline matrices, say Bézier and Catmull-Rom? Both are cubics, so you can represent the other curve exactly with new control points G2 G 1 B 1 T(t) G 2 B 2 T(t) G 2 =??? CS-C3100 Fall 2017 Lehtinen 53

54 Change of Basis, Other Direction Given B1...B4, how to get back to canonical 1, t, t 2, t 3? B 1 (t) B 2 (t) B 3 (t) B 4 (t) = B t t 2 t 3 CS-C3100 Fall 2017 Lehtinen 54

55 Change of Basis, Other Direction That s right, with the inverse matrix! Given B1...B4, how to get back to canonical 1, t, t 2, t 3? 1 t t 2 t 3 = B /3 2/ / B 1 (t) B 2 (t) B 3 (t) B 4 (t) CS-C3100 Fall 2017 Lehtinen 55

56 Recap Cubic polynomials form a vector space. Bernstein basis is canonical for Bézier. Can be seen as influence function of data points Or data points are coordinates of the curve in the Bernstein basis We can change between basis with matrices. CS-C3100 Fall 2017 Lehtinen 56

57 Questions? CS-C3100 Fall 2017 Lehtinen 57

58 A Cubic Only Gets You So Far What if you want more control? CS-C3100 Fall 2017 Lehtinen 58

59 CS-C3100 Fall 2017 Lehtinen 59 Higher-Order Bézier Curves > 4 control points Bernstein Polynomials as the basis functions For polynomial of order n, the i th basis function is Every control point affects the entire curve Not simply a local effect More difficult to control for modeling You will not need this in this class

60 Subdivision of a Bézier curve Can we split a Bezier curve into two in the middle, using two new Bézier curves? Would be useful for adding detail, as a single cubic doesn t get you very far, and higher-order curves are nasty.? CS-C3100 Fall 2017 Lehtinen 60

61 Subdivision of a Bezier curve Can we split a Bezier curve into two in the middle, using two new Bézier curves? The resulting curves are again a cubic (Why?)? CS-C3100 Fall 2017 Lehtinen

62 Subdivision of a Bezier curve Can we split a Bezier curve into two in the middle, using two new Bézier curves? The resulting curves are again a cubic (Why? A cubic in t is also a cubic in 2t) (Why? a 0 (2t) 3 =8a 0 t 3, etc.) Hence it must be representable using the Bernstein t1=2t t2=2t-1 basis. So yes, we can!? cubic t=0 t=0.5 t=1 CS-C3100 Fall 2017 Lehtinen 62

63 Subdivision of a Bezier curve Can we split a Bezier curve into two in the middle, using two new Bézier curves? The resulting curves are again a cubic (Why? A cubic in t is also a cubic in 2t) Hence it must be representable using the Bernstein basis. So yes, we can!? t1=2t t2=2t-1 cubic t=0 t=0.5 t=1 CS-C3100 Fall 2017 Lehtinen 63

64 De Casteljau Construction Take the middle point of each of the 3 segments Construct the two segments joining them Take the middle of those two new segments Join them Take the middle point P P 2 P 1 P 2 P 1 P P 3 CS-C3100 Fall 2017 Lehtinen 64

65 Result of Split in Middle The two new curves are defined by P1, P 1, P 1, and P P, P 2, P 3, and P4 Together they exactly replicate the original curve! Originally 4 control points, now 7 (more control) P 2 P 1 P 2 P 1 P P 3 P4 P1 CS-C3100 Fall 2017 Lehtinen 65

66 Sanity Check Do we get the middle point? B 1 (t)=(1-t) 3 B 2 (t)=3t(1-t) 2 B 3 (t)=3t 2 (1-t) B 4 (t)=t 3 P 1 P 1 P 2 P 2 P P 3 P 1 =0.5(P 1 + P 2 ) P 2 =0.5(P 2 + P 3 ) P 3 =0.5(P 3 + P 4 ) P 1 =0.5(P 1 + P 2 ) P 2 =0.5(P 2 + P 3 ) P =0.5(P 1 + P 2 ) =0.5 (0.5(P 1 + P 2 )+0.5(P 2 + P 3 )) =0.5 (0.5 [0.5(P 1 + P 2 )+0.5(P 2 + P 3 )] [0.5(P 2 + P 3 )+0.5(P 3 + P 4 )] =1/8P 1 +3/8P 2 +3/8P 3 +1/8P 4 CS-C3100 Fall 2017 Lehtinen 66

67 De Casteljau Construction Actually works to construct a point at any t, not just 0.5 Just subdivide the segments with ratio (1-t), t (not in the middle) t t t t t t CS-C3100 Fall 2017 Lehtinen 67

68 De Casteljau Construction Actually works to construct a point at any t, not just 0.5 Just subdivide the segments with ratio (1-t), t (not in the middle) 1-t t t t t 1-t t t 1-t CS-C3100 Fall 2017 Lehtinen 68

69 Questions? CS-C3100 Fall 2017 Lehtinen 69

70 Aside: Extra Credit Suggestion TrueType font glyphs (the images of the characters) consist of quadratic Bézier curves chained together. Why don t you write code that loads the glyphs using the Win32 API, and either Convert them to cubic Bézier and display them using the drawing code you already wrote Or write code for drawing quadratic curves directly. See this MSDN documentation page for a starting point! CS-C3100 Fall 2017 Lehtinen 70

71 Linear Transformations & Cubics What if we want to transform each point on the curve with a linear transformation M? P (t)= M CS-C3100 Fall 2017 Lehtinen 71

72 Linear Transformations & Cubics What if we want to transform each point on the curve with a linear transformation M? Because everything is linear, it s the same as transforming the only the control points P (t)= M = M CS-C3100 Fall 2017 Lehtinen 72

73 Linear Transformations & Cubics Homogeneous coordinates also work Means you can translate, rotate, shear, etc. Also, changing w gives a tension parameter Note though that you need to normalize P by 1/w P (t)= M = M CS-C3100 Fall 2017 Lehtinen 73

74 Differential properties of curves Motivation Compute normal for surfaces Compute velocity for animation Analyze smoothness CS-C3100 Fall 2017 Lehtinen 74

75 Velocity First derivative w.r.t. t Can you compute this for Bezier curves? P(t) = (1-t) 3 P 1 + 3t(1-t) 2 P 2 + 3t 2 (1-t) P 3 + t 3 P 4 You know how to differentiate polynomials... CS-C3100 Fall 2017 Lehtinen 75

76 Velocity First derivative w.r.t. t Can you compute this for Bezier curves? P(t) = (1-t) 3 P 1 + 3t(1-t) 2 P 2 + 3t 2 (1-t) P 3 + t 3 P 4 P (t) = -3(1-t) 2 P1 + [3(1-t) 2-6t(1-t)] P2 + [6t(1-t)-3t 2 ] P3 + 3t 2 P4 CS-C3100 Fall 2017 Lehtinen Sanity check: t=0; t=1 Can also write this using a matrix B try it out! 76

77 Tangent The tangent to the curve P(t) can be defined as T(t)=P (t)/ P (t) normalized velocity, T(t) = 1 This provides us with one orientation for swept surfaces in a little while CS-C3100 Fall 2017 Lehtinen 77

78 Questions? CS-C3100 Fall 2017 Lehtinen 78

79 Curvature Derivative of unit tangent K(t)=T (t) Magnitude K(t) is constant for a circle Zero for a straight line Always orthogonal to tangent, ie. Can you prove this? (Hints: T(t) =1, (x(t)y(t)) =?) T(t) Low curvature K T =0 T(t) T(t) K(t) K(t) K(t) High curvature CS-C3100 Fall 2017 Lehtinen 79

80 Geometric Interpretation K is zero for a line, constant for circle What constant? 1/r 1/ K(t) is the radius of the circle that touches P(t) at t and has the same curvature as the curve T(t) r=1/k T(t) r=1/k K(t) K(t) CS-C3100 Fall 2017 Lehtinen 80

81 Curve Normal Normalized curvature: T (t)/ T (t) N(t) Low curvature N(t) N(t) High curvature CS-C3100 Fall 2017 Lehtinen 45 81

82 Questions? CS-C3100 Fall 2017 Lehtinen 82

83 Orders of Continuity C 0 = continuous The seam can be a sharp kink G 1 = geometric continuity Tangents point to the same direction at the seam C 1 = parametric continuity Tangents are the same at the seam, implies G 1 C 2 = curvature continuity Tangents and their derivatives are the same CS-C3100 Fall 2017 Lehtinen C 0 G 1 C 1 83

84 Connecting Cubic Bézier Curves How can we guarantee C 0 continuity? How can we guarantee G 1 continuity? How can we guarantee C 1 continuity? C 2 and above gets difficult CS-C3100 Fall 2017 Lehtinen 84

85 Connecting Cubic Bézier Curves Where is this curve C 0 continuous? G 1 continuous? C 1 continuous? What s the relationship between: the # of control points, and the # of cubic Bézier subcurves? CS-C3100 Fall 2017 Lehtinen 85

86 Questions? CS-C3100 Fall 2017 Lehtinen 86

87 Cubic B-Splines 4 control points Locally cubic Cubics chained together, again. CS-C3100 Fall 2017 Lehtinen 87

88 Cubic B-Splines 4 control points Locally cubic Cubics chained together, again.. BUT with a sliding window of 4 control points! CS-C3100 Fall 2017 Lehtinen 88

89 Cubic B-Splines 4 control points Locally cubic Cubics chained together, again. BUT with a sliding window of 4 control points! CS-C3100 Fall 2017 Lehtinen 89

90 Cubic B-Splines 4 control points Locally cubic Cubics chained together, again. BUT with a sliding window of 4 control points! CS-C3100 Fall 2017 Lehtinen 90

91 Cubic B-Splines 4 control points Locally cubic Cubics chained together, again. Curve is not constrained to pass through any control points A BSpline curve is also bounded by the convex hull of its control points. CS-C3100 Fall 2017 Lehtinen 91

92 Cubic B-Splines: Basis These sum to 1, too! B2 B3 B1 B4 B 1 (t) = 1 6 (1 t)3 B 2 (t) = 1 6 (3t3 6t 2 + 4) B 3 (t) = 1 6 ( 3t3 +3t 2 +3t + 1) B 4 (t) = 1 6 t3 CS-C3100 Fall 2017 Lehtinen 92

93 Cubic B-Splines: Basis CS-C3100 Fall 2017 Lehtinen 93

94 Cubic B-Splines Local control (windowing) Automatically C 2, and no need to match tangents! CS-C3100 Fall 2017 Lehtinen 94

95 B-Spline Curve Control Points Default BSpline BSpline with derivative discontinuity Repeat interior control point BSpline which passes through end points Repeat end points CS-C3100 Fall 2017 Lehtinen 95

96 Bézier B-Spline Bézier BSpline But both are cubics, so one can be converted into the other! CS-C3100 Fall 2017 Lehtinen 96

97 Easy Brain Teaser Can you derive formulas for conversion between two different cubic spline matrices, say Bézier and Catmull-Rom? Both are cubics, so you can represent the other curve exactly with new control points G2 G 1 B 1 T(t) G 2 B 2 T(t) G 2 =??? CS-C3100 Fall 2017 Lehtinen 97

98 Solution G 1 B 1 T(t) G 2 B 2 T(t) (This must hold for all t!) CS-C3100 Fall 2017 Lehtinen

99 Solution G 1 B 1 T(t) G 2 B 2 T(t) (This must hold for all t!) CS-C3100 Fall 2017 Lehtinen

100 Converting between Bézier & BSpline Simple with the basis matrices! Note that this only works for a single segment of 4 control points P(t) = G B1 T(t) = G B1 (B2-1 B2) T(t)= (G B1 B2-1 ) B2 T(t) G B1 B2-1 are the control points for the segment in new basis CS-C3100 Fall 2017 Lehtinen 100

101 Converting between Bézier & BSpline original control points as Bézier new BSpline control points to match Bézier new Bézier control points to match BSpline original control points as BSpline CS-C3100 Fall 2017 Lehtinen 101

102 Why Bother with B-Splines? Automatic C 2 is nice! Also, B-Splines can be split into segments of nonuniform length without affecting the global parametrization. Non-uniform B-Splines We ll not do this, but just so you know. CS-C3100 Fall 2017 Lehtinen 102

103 NURBS (Generalized B-Splines) Rational cubics Use homogeneous coordinates, just add w! Provides a tension parameter to control points NURBS: Non-Uniform Rational B-Spline non-uniform = different spacing between the blending functions, a.k.a. knots rational = ratio of cubic polynomials (instead of just cubic) implemented by adding the homogeneous coordinate w into the control points. CS-C3100 Fall 2017 Lehtinen 103

104 Questions? CS-C3100 Fall 2017 Lehtinen 104

105 That s All for Today, Folks Fun stuff to know about function/vector spaces Inkscape is an open source vector drawing program for Mac/Windows. Try it out! vectorportal.com CS-C3100 Fall 2017 Lehtinen 105